We develop a dynamic zoning strategy for China's data centers that coordinates computing, electricity, and heat systems to identify regional mitigation priorities and reduce carbon emissions.
Rapid advances in artificial intelligence (AI) are increasing data center electricity demand and intensifying the challenge of decarbonizing the sector, but it remains unclear how operational net zero CO₂ emissions can be achieved at national and regional scales. We develop a provincial framework for China’s data center sector from 2024 to 2050 that separates exogenous demand scenarios from mitigation pathways and integrates facility efficiency, power grid decarbonization, computing allocation and waste heat recovery (WHR) within coupled electricity–heat accounting. Electricity consumption in 2050 ranges from 372.6 to 2,255.8 TWh. Under the Ambition pathway, four of five demand scenarios reach operational net zero by 2050, while High Power retains 35.6 Mt CO₂. In the Mid-case, grid decarbonization, PUE improvement and WHR contribute 60.5%, 24.7% and 14.5% of cumulative avoided emissions. No scenario reaches net zero without WHR, but its contribution is limited by local heat demand. East Data West Computing allocation adds 32.7 Mt CO₂ during 2024–2040 before western grids decarbonize. To guide the siting of new computing capacity and the priority measures for each region, we classify provinces into four dynamic zones based on electricity related CO₂ intensity per unit computing and effective waste heat utilization capacity. Priority provinces fall from 16 in 2030 to 13 in 2050, with changing membership as heat constraints emerge. These findings show that net zero for the data center sector requires zoning management that coordinates computing, power and heat systems.
Wei Li is a Professor at Beijing Normal University, China. His research focuses on environmental impact assessment, biodiversity valuation, ecological economics, and green finance.